JP3747957B2 - Connection table editing device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention provides a connection table for morpheme analysis that holds a connection word related to whether or not a front word or part of speech, a back word or part of speech, and a connection of a front word or part of speech and a back word or part of speech can be connected. The present invention relates to a connection table editing apparatus that performs editing on a network.
[0002]
[Prior art]
In a device that processes a natural language sentence such as a Japanese sentence (for example, a machine translation device, a question and answer device, a computer-aided education device, etc.), a morphological analysis is first performed on the natural language sentence, and then a desired operation is performed. To do.
[0003]
Conventionally, a morphological analyzer for Japanese sentences that performs such morphological analysis is composed of a morphological analysis unit (morpheme analysis program unit), a Japanese dictionary, a utilization ending table, and a connection table (by part of speech). Have been proposed (see Japanese Patent Publication No. 05-52543).
[0004]
In addition to the connection table indicating whether or not connection is possible, a connection table using a connection weight table that can take continuous values has been proposed (see Japanese Patent Application Laid-Open No. 05-13327).
[0005]
The data in the connection table is described / changed according to the connection rules that define the connection relationship (literatures “Yuji Matsumoto, Ikuo Kurohashi, Takehito Utsuro, Hiroshi Myoki, Makoto Nagao,“ Japanese Morphological Analysis System JUMAN Instruction Manual version ” 2.0 ", February 1993]).
[0006]
[Problems to be solved by the invention]
In order to change the output result in the morphological analyzer, it is conceivable to modify the dictionary or the connection table.
[0007]
Conventionally, it was rarely assumed that the user freely changed the dictionary or connection table, that is, customized, so as shown below, the connection table editing that satisfies the requirements necessary especially for correcting the connection table The device did not exist.
[0008]
In a conventional connection table editing device, one rule (or number) is added to each line when adding or modifying values or items in the connection table. Le Edit), it was difficult to refer to the relationship with other data. Hereinafter, a format in which one rule or number rule is described in one line is referred to as a rule format.
For example, FIG. 2 shows an example of connection weights between parts of speech described in the form of one rule per line. In FIG. 2, the front part indicates the previous part of speech when there are two connectable parts of speech, and the rear part indicates the subsequent part of speech when there are two connectable parts of speech. Taking a line in which the part is a noun and the trailing part is a case particle, for example, it means that the connection weight between both parts of speech (noun and case particle) is 0.8. Here, it is assumed that the connection weight ranges from 0 to 1, and that the larger the numerical value, the higher the probability that both parts of speech are connected.
[0009]
In the conventional connection table editing apparatus, some or all of the data in the format shown in FIG. 2 is displayed in the rule format for editing.
[0010]
Here, when trying to change the connection weight between a noun and a verb, I would like to refer to the connection weight between a case particle and a verb, but it also happens that the connection weight rule between a case particle and a verb exists quite backward. obtain. In addition, when it is desired to simultaneously compare three rules with reference to the connection weights of adjunct and verb, it is very difficult with the conventional connection table editing apparatus.
[0011]
Also, when adding items such as new parts of speech, search for connection rules for existing parts of speech to find differences from the desired connection rules, copy appropriate parts, and copy some if necessary. It is desirable to take the procedure of correcting. Furthermore, in order to reduce the amount of data, it is possible to check whether a newly created rule group overlaps with any existing part-of-speech connection rule group and combine the overlapping parts into one.
[0012]
For example, suppose that it is desired to add a new “sa-changing noun” to an existing connection table. At this time, it is necessary to create a connection rule for all the parts of speech currently defined and the saun noun. All roux Le Since it takes an enormous amount of time to create from scratch, a rule group having a small difference with respect to a desired connection condition is searched from among a group of existing part-of-speech connection rules. The sa noun has almost the same connection condition as the noun, but it is thought that only the connection weight with the verb “s” is different, so copy all the connection rules of the noun and copy only the connection weight with the verb “s”. change. It is checked whether or not the connection rule group of the sa-variable noun created here matches the rule group of another part of speech that already exists. If it matches the existing rule group, it is not necessary to create a new connection rule group for the sub-noun, and the existing data can be used as it is. However, with the conventional connection table editing device, it is difficult to see the relationship with other parts at a glance when describing or correcting the connection table, and it is difficult to compare rules and check for duplication. Editing such a connection table requires a great deal of effort.
[0013]
[Means for Solving the Problems]
The invention of claim 1 is for morphological analysis that retains a front word or part of speech, a rear word or part of speech, and a connection rule relating to whether or not the front word or part of speech can be connected to the rear word or part of speech. A connection table editing apparatus for editing a connection table includes the following means.
[0014]
That is, For morphological analysis Based on the information reading means for reading information stored in the connection table and the information read by the information reading means, the front word or part of speech is expanded into a row or column heading. As well as Expand backward word or part of speech to column or row heading And each A table creation means that creates a table composed of rows and columns by arranging a connection rule relating to whether or not a front word or part of speech and a back word or part of speech can be connected at the intersection of a row and a column, and a table creation means Display means for displaying a table composed of rows and columns, input receiving means for receiving input from the user, Based on the information from the input accepting means, the edit operation means for editing the table created by the table creation means, and after editing the table by the edit operation means, the connection rule group possessed by the row or column related to the edit location is another Duplicate check means for determining whether or not a connection rule group included in a row or a column overlaps, and a connection rule group included in another row or column included in a connection rule group included in the row or column related to the edited portion by the overlap check means A merge means for adding another duplicate row or column heading to one row or column heading and deleting the other row or column when it is determined that the other row or column is duplicated , A table with rows and columns For morphological analysis Format converting means for converting to a format corresponding to the connection table, and information writing means for writing the information after the format conversion to the connection table And comprising .
[0015]
In the morphological analyzer of the invention of claim 1, the information reading means reads the information stored in the connection table, and the table creating means reads the previous word or part of speech based on the information read by the information reading means. Expands to a row or column heading, expands a backward word or part of speech to a column or row heading, and determines whether or not a front word or part of speech and a backward word or part of speech can be connected at the intersection of the row and column Arrange the connection rules, create a table of rows and columns, display the table of rows and columns, input accepting means accepts input from the user, edit operation means accepts input After editing the table consisting of rows and columns based on the information from the means, the format conversion means converts the table consisting of rows and columns into a format corresponding to the connection table, and after the information writing means converts the format Information connection table Export.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
(First embodiment)
FIG. 3 shows a functional configuration of the morpheme analyzer including the connection table editing device according to the first embodiment. Actually, the morpheme analyzer according to the first embodiment has a large capacity. This is realized by a computer system including the auxiliary storage device, and the hardware configuration thereof is omitted. The morpheme analyzer in the first embodiment is functionally composed of a morpheme analyzer 31, a connection table 32, a dictionary 33, and a connection table editor 34 as shown in FIG.
[0017]
The connection table 32 holds values that determine whether or not words (parts of speech) can be connected or the connection weights. The connection table 32 is edited by the connection table editing device 34 and referred to by the morphological analysis unit 31. Although there may be a plurality of connection tables 31, it is assumed that only one connection table 31 exists in the first embodiment.
[0018]
The dictionary 33 holds information on words expressed by combinations of headings and parts of speech, and is referred to by the morpheme analysis unit 31. Although there may be a plurality of dictionaries 33, it is assumed that only one dictionary 33 exists in the first embodiment.
[0019]
The morpheme analysis unit 31 reads a character string and performs morpheme analysis on the character string with reference to information stored in the connection table 32 and the dictionary 33. That is, in the case where a Japanese sentence is taken as an example, the morphological analysis unit 31 searches the unprocessed portion of the input character string for a matching word (group) by referring to the dictionary 33, and uses it at that time. When a word comes out, a utilization ending is added using utilization ending information in the dictionary 33 or in the morpheme analysis unit 31, and whether the searched word can be connected to the front and rear morpheme candidates. Is determined by referring to the connection table 32, and if the connection is possible, the word is output as a morpheme candidate. If the connection is impossible, the same processing is performed on the other word candidates, and finally the word is obtained. The morpheme analysis result is output.
[0020]
The connection table editing device 34 accepts user input, displays the rule format data stored in the connection table 32 in a table format, accepts the edit, and if there is a change, the table format data is converted into a rule format. The stored information in the edit table 32 is changed by converting to
[0021]
As shown in FIG. 1, the connection table editing device 34 includes a reading / writing unit 11, a conversion unit 12, a data management unit 13, a display unit 14, and an input receiving unit 15. Also, the same and corresponding parts as those in FIG.
[0022]
The read / write unit 11 reads the rule format data stored in the connection table 33 according to an instruction from the data management unit 13 and sends the rule format data to the conversion unit 12. The read / write unit 11 receives the rule format data from the conversion unit 12 and receives the rule format data. Data is written to the connection table 33.
[0023]
The conversion unit 12 receives the data in the rule format from the reading / writing unit 11 and arranges it according to a predetermined rule, converts it into a table format and sends it to the data management unit 13, and also sends the data in the table format from the data management unit 13. Data is received, converted into a rule format, and sent to the reading / writing unit 11.
[0024]
The display unit 14 receives tabular data sent from the data management unit 13 and displays it in a tabular format on a display device (not shown).
[0025]
The input receiving unit 15 receives user input from a user interface device (not shown), analyzes the input content, and sends an instruction to the data management unit 13 based on the analysis result.
[0026]
The data management unit 13 sends the tabular data received from the conversion unit 12 to the display unit 14, receives an instruction from the input receiving unit 15, performs desired editing on the tabular data, and edits the edited table. The format data is sent to the display unit 14, and after the editing is completed, the tabular data is sent to the conversion unit 12.
[0027]
4, the data management unit 13 includes a control unit 41, a search unit 42, an edit protection unit 43, a duplication check unit 44, a merge unit 45, a copy unit 46, a comparison unit 47, and a calculation unit 48. Consists of. Also, the same and corresponding parts as those in FIG.
[0028]
The control unit 41 receives tabular data from the conversion unit 12, sends the tabular data to the display unit 14, receives an instruction from the input receiving unit, and receives the instructions from the units 42 to 42 in the data management unit 13 as necessary. 48 is used to edit the tabular data. The search unit 42 searches for desired data from tabular data in response to an instruction from the control unit 41. In response to an instruction from the control unit 41, the edit protection unit 43 sets only the row or column data including the data searched by the search unit 42 to be editable, and sets other data to be uneditable. It is. The duplication check unit 44 receives an instruction from the control unit 41 and checks whether or not the corrected row or column data overlaps with existing row or column data every time the user edits. Yes, and notifies the control unit 41 of the result. The merging unit 45 merges the corrected data with the existing data in response to an instruction from the control unit 41 after confirmation by the user when there is duplication in the result data examined by the duplication check unit 44. In response to an instruction from the control unit 41, the copying unit 46 copies the specified portion of data to another location in the tabular data. The comparison unit 47 receives an instruction from the control unit 41, examines differences in data of a plurality of rows or columns in tabular data, and notifies the control unit 41 of the differences. The calculation unit 48 performs numerical calculation in response to an instruction from the control unit 41 and notifies the control unit 41 of the calculation result.
[0029]
The operation of the connection table editing device 34 according to the first embodiment will be described below.
[0030]
First, the basic operation of the connection table editing apparatus 34 in the first embodiment will be described by dividing it into three parts: display, editing, and writing.
[0031]
First, the operation when the connection table editing device 34 reads and displays data from the connection table 32 will be described. The input receiving unit 15 receives an input of a display instruction from the user, analyzes the input content, obtains that the input content is a display instruction, sends the display instruction to the data management unit 13, and receives the display instruction The data management unit 13 sends a data read instruction to the reading / writing unit 11, and the reading / writing unit 11 reads the data (rule format as shown in FIG. 2) stored in the connection table 32 and converts the information into the conversion unit 12. The conversion unit 12 converts the data in the rule format into tabular data and sends it to the data management unit 13. The data management unit 13 sends the data in the table format to the display unit 14, and the display unit 14 manages the data. The tabular data from the unit 13 is displayed in tabular format on a display device (not shown).
[0032]
FIG. 5 shows an example of tabular information displayed on a display device (not shown). In FIG. 5, the front shows the front part and the back shows the rear part. The number displayed at the intersection between each part of speech is the connection weight between both parts of speech.
[0033]
Next, the operation when the connection table editing device 34 edits the data in the table format (format in FIG. 5) will be described. The user issues an editing instruction for the data displayed in the table format, and the input receiving unit 15 that receives the editing instruction input analyzes the input content and sends a desired instruction to the data management unit 13. The desired editing is performed on the data in the format, and the edited data is sent to the display unit 14. The display unit 14 displays the data in a table format on a display device (not shown).
[0034]
Further, the operation when the connection table editing device 34 writes the edited data (format in FIG. 5) in the connection table 32 (format in FIG. 2) will be described. After the editing operation is completed, the user inputs a writing instruction, the input receiving unit 15 analyzes the input content and sends the writing instruction to the data management unit 13, and the data management unit 13 converts the tabular data after editing. The conversion unit 12 converts the tabular data into the rule format, sends the rule format data to the reading / writing unit 11, and the reading / writing unit 11 writes the rule format data to the connection table 32 (deletion).・ Change / Add).
[0035]
Although the basic operation is as described above, a more detailed operation will be described below. First, referring to the flowchart of FIG. 6, the operation when a series of processing of search, edit protection, correction, and duplication check is performed on tabular data after data as shown in FIG. 5 is displayed. explain.
[0036]
In response to an input meaning “search for connection between nouns and verbs” from the user (which may be in a form such as “search for a place where the value of the noun and connection is 0.6”), the input receiving unit 15 And an instruction corresponding to “search for connection between nouns and verbs” is instructed to the data management unit 13, and the calculation unit 14 in the data management unit 13 receiving the instruction receives the instruction and sends it to the search unit 42. The search unit 42 searches for the data of “noun and verb connection” from the entire tabular data, sends the search result to the control unit 41, and the control unit 41 instructs the edit protection unit 43. The edit protection unit 43 sets only the searched row or column data to be editable and sets other data to be uneditable, and then returns control to the control unit 41. Is sent to the display unit 14, and the display unit displays the searched position first. To display the data in the (step 601, 602).
The data displayed at this time is as shown in FIG. In FIG. 7, the front shows the front part, and the back shows the rear part. The number displayed at the intersection between each part of speech is the connection weight between both parts of speech.
[0037]
If the user refers to another part again, the process returns to step 601, and the search and edit protection operations are performed again. If not, the process proceeds to the subsequent step (step 603).
[0038]
Subsequently, as described above, the tabular data is edited (corrected), and then it is checked whether the corrected row or column data is duplicated with the existing data (details of the duplicate check will be described later). ) When the user re-edits, the correction and duplication check operations are repeated (steps 604 to 606).
[0039]
The operation for correcting the existing data is as described above. The details of the duplication check operation in step 605 will be described with reference to the flowchart of FIG.
[0040]
The control unit 41 instructs the duplication check unit 44 to perform duplication check, and in response to this, the duplication check unit 44 checks whether the corrected row or column data is duplicated with existing data. If there is an overlap, the control unit 41 informs the display unit 14 to display the overlap portion, and the display unit 14 sends the data with the overlap portion as the head. It is displayed (steps 801 and 802).
When a user who has viewed duplicates makes an input that means "identify duplicate data", delete the corrected row or column data in the duplicate data, and use the deleted part of speech as duplicate data. Is added to the part of speech of the existing data (steps 803 and 804).
[0041]
A display example when displaying in step 802 is shown in FIG. In FIG. 9, the front shows the front part and the back shows the rear part. The number displayed at the intersection between each part of speech is the connection weight between both parts of speech. In FIG. 9, it is assumed that the noun and the data of the special noun 1 are duplicated in the part of speech of the front part, and the noun line is corrected.
[0042]
FIG. 10 shows a display example after the data is made identical in step 804 for the duplication as shown in FIG. Also in FIG. 10, the front part shows the front part and the rear part shows the rear part. The number displayed at the intersection between each part of speech is the connection weight between both parts of speech. In FIG. 10, the noun data of the part of speech of the front part is deleted, and the deleted part of speech (noun) is added to the special noun 1.
[0043]
In this way, the data amount stored in the connection table 32 can be reduced by making the data of the duplicated rows or columns identical.
[0044]
The data stored in the connection table 32 in the present embodiment includes data in the form of one line and one rule as shown in FIG. 2 and data in the form of several rules written in one line. Furthermore, as already mentioned, there are some data that are identified when the data in the row or column is duplicated.
[0045]
FIG. 11 shows an example of the data in the number rule format in one line stored in the connection table 32. In FIG. 11, the noun suffix and the noun noun suffix of the noun indicate that the connection weight with the verb is 0.5, and the noun form indicator, the determinator and the special This indicates that the connection weight value of the punctuation marks is not entered (omitted).
[0046]
Next, details of the operation of the conversion unit 12 that has received the data stored in the connection table 32 via the read / write unit 11 will be described with reference to the flowchart of FIG.
[0047]
The conversion unit 12 receives the data in the rule format from the reading / writing unit 11, extracts the front part and the rear part, and arranges them (so that the partial character strings are arranged close to each other). A partial list and a trailing partial list are created, the size of the table is determined from the number of items in the list, and empty tabular data containing no value is created (steps 1201 and 1202).
[0048]
Subsequently, the connection weight value is extracted from the data in the rule format using the heading part (part of speech) in the front part list and the back part list as a key, and the value (connection weight) is embedded in the tabular data. . At this time, the default value is embedded for those for which no value is entered (step 1203).
[0049]
The parts of text that contain the same substrings, such as two parts of speech, such as a noun (generally noun) and a noun sa-variant noun, are arranged close to each other. Mutual This is because it can be considered that the connection weights are very similar to each other, and it can be easily corrected while comparing parts of speech.
[0050]
The operation when a new part of speech is added to the connection table 32 after the tabular data is displayed on a display device (not shown) will be described with reference to the flowchart of FIG.
[0051]
Upon receiving an instruction to add a new part of speech from the user, the input receiving unit 15 analyzes the input content and sends the instruction to the control unit 41 in the data management unit 13 so that the area of the row or column to be added is displayed in a tabular format. The data is secured in the data and only the part-of-speech heading is entered, the connection weight portion is left blank, and the data is sent to the display unit 14. The display unit displays the data portion of the added row or column (step 1301).
[0052]
FIG. 14 shows a display example when the processing of step 1301 is performed. In FIG. 14, the front shows the front part and the rear shows the rear part. The number displayed at the intersection between each part of speech is the connection weight between both parts of speech. In FIG. 14, a newly added heading (sa variable noun) is displayed, and the connection weight is left blank (“-” in FIG. 14 indicates a blank).
[0053]
Next, search processing similar to the processing in steps 601 to 603 in the flowchart of FIG. 6 is performed. However, the search location is not necessarily displayed at the top of the display, but can be displayed at an arbitrary location (step 1302).
[0054]
In step 1302, when the user receives an instruction to search for a plurality of locations and further display the difference between the data of the plurality of search locations, the input receiving unit 15 analyzes the input content and instructs the display of the differences. To the control unit 41 in the data management unit 13, the control unit 41 controls the comparison unit 47, the comparison unit 47 compares the data of a plurality of search locations, and notifies the comparison result to the control unit 41, The control unit 41 sends the tabular data together with the comparison result to the display unit 14, and the display unit 14 displays the difference between the data in addition to the data of the search location (steps 1303 and 1304).
[0055]
When the user refers to the difference between the data and the user gives an instruction to copy the data from the data at the search location to the newly added row or column, the input receiving unit 15 analyzes the input content and performs the instruction. To the control unit 41, the control unit 41 controls the copying unit 46, and the copying unit copies the specified portion of the data of the search location to the specified portion of the newly added row or column, and controls The unit 41 sends the tabular data after copying to the display unit 14, and the display unit 14 displays the data in tabular format (step 1305).
[0056]
Thereafter, the same processing as steps 604 to 605 in the flowchart of FIG. 6 is performed to perform data correction processing and post-correction duplication check processing (step 1306).
[0057]
As described above, according to the first embodiment, the reading / writing unit 11 Contact The rule format data stored in the connection table 32 is read out, and the table conversion unit 11 is read out. Is Data is converted into tabular data consisting of rows and columns, the data management unit 13 receives the data and passes it to the display unit 14, and the display unit 14 displays the tabular data and When the input is received, the input receiving unit 15 analyzes the input contents and notifies the data management unit 13 of an instruction of the analysis result, and the data management unit performs desired editing in accordance with the instruction, and after editing When the data is passed to the display unit 14 and the display unit 14 displays the edited data in a tabular format and the user gives an instruction to write, the input receiving unit 15 analyzes the input content and instructs the analysis result Is sent to the data management unit 13, the data management unit 13 passes the tabular data to the conversion unit 12, the conversion unit 12 converts the tabular data into rule format data, and the reading / writing unit 11 Rule form Since such writing over data in the connection table 32, can display a large amount of information at a time on the display device, the reference and edit data can be performed easily.
[0058]
Further, since the reading / writing unit 11 arranges the data, data having connection weights similar to each other can be displayed nearby, and editing can be performed while referring to data having similar connection weights. Can do.
[0059]
In addition, the search unit 42 searches for the designated location, and the edit protection unit 43 sets only the search location to be editable and sets other portions to be uneditable. Can be displayed automatically, and erroneous correction can be prevented.
[0060]
In addition, after the data correction, the duplication check unit 44 examines the duplication between the corrected data and the existing data, and when there is duplication, the data merged by the merging unit 45 and the existing data are merged into one. Thus, correction errors can be prevented and the amount of data stored in the display and connection table 32 can be reduced.
[0061]
Furthermore, since the comparison unit 47 examines the difference between the reference data and the user considers the difference, the information copying unit 46 copies an arbitrary part from the reference data to another place. Time required for data input can be greatly reduced.
[0062]
(Other embodiments)
In the first embodiment, the rule format data is stored in the connection table. However, the present invention is applicable if data other than the table format is stored in the connection table.
[0063]
In the first embodiment, the data is aligned by the conversion unit. However, the data management unit is provided with an alignment unit for aligning data according to a predetermined rule. You may make it align.
[0064]
Furthermore, in the first embodiment, the copying unit copies data when a new part of speech is added. However, the copying unit may copy data at an arbitrary timing according to an instruction from the user. Similarly, the edit protection, duplication check, and merge and compare processes may be performed at arbitrary timings.
[0065]
In the first embodiment, the application to Japanese is shown, but the present invention can also be applied to other languages.
[0066]
【The invention's effect】
As described above, according to the present invention, Based on the information read out from the information read in the connection table for morphological analysis and the information read out by the information reading means, the front word or part of speech is expanded into a row or column heading and the rear word or part of speech is Expands to column or row headings, and creates a table of rows and columns by placing connection rules related to the connection between the front word or part of speech and the back word or part of speech at the intersection of each row and column. Based on information from the table creation means, a display means for displaying a table composed of rows and columns created by the table creation means, an input acceptance means for accepting input from the user, and a table creation means. Editing operation means for editing the created table, and after editing the table by the editing operation means, whether or not the connection rule group possessed by the row or column related to the edited location overlaps with the connection rule group possessed by another row or column When the duplication check means and the duplication check means judge that the connection rule group possessed by the row or column related to the edited location overlaps with the connection rule group possessed by another row or column, the duplicate rows or columns A merge means for adding another duplicate row or column heading to one row or column heading of the columns and deleting the other row or column, and a table composed of the rows and columns to the connection table for morphological analysis Since it is provided with format conversion means for converting to a corresponding format and information writing means for writing information after the format conversion to the connection table. A large amount of information can be displayed systematically at once, Since you can compare rules and check duplicates, It becomes easy to refer to and edit information, thereby reducing erroneous editing and greatly shortening the time required for editing.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a functional configuration of a connection table editing apparatus according to a first embodiment.
FIG. 2 is a diagram showing an example of connection weights between parts of speech described in the form of one line and one rule.
FIG. 3 is a block diagram showing a functional configuration of the morphological analyzer according to the first embodiment.
FIG. 4 is a block diagram illustrating a functional configuration of a calculation unit according to the first embodiment.
FIG. 5 is a diagram showing a display example of a display unit in the first embodiment.
FIG. 6 is a flowchart showing a flow of a series of processing of search, edit protection, correction, and duplication check in the first embodiment.
FIG. 7 is a diagram showing a display example of the display unit after searching in the first embodiment.
FIG. 8 is a flowchart showing a flow of a series of processing of duplication check and data identification in the first embodiment.
FIG. 9 is a diagram showing a display example of the display unit after the duplication check in the first embodiment.
FIG. 10 is a diagram showing a display example of the display unit after data is made identical in the first embodiment.
FIG. 11 is a diagram showing an example of connection weights between parts of speech described in the form of one line number rule.
FIG. 12 is a flowchart illustrating a processing flow of a conversion unit according to the first embodiment.
FIG. 13 is a flowchart showing the flow of processing when adding new data in the first embodiment.
FIG. 14 is a diagram showing a display example of new data in the first embodiment.
[Explanation of symbols]
11 Reading / writing section
12 Conversion unit
13 Data Management Department
14 Display section
15 Input receiving part
32 Connection table
34 Connection table editing device
41 Control unit
42 Search unit
43 Editing Protection Department
44 Duplication check part
45 Merger
46 Copying Department
47 comparison part
48 Calculation unit

Claims (4)

Edit the connection table for morphological analysis that holds the front word or part of speech, the rear word or part of speech, and the connection rule relating to the possibility of connection of the front word or part of speech and the rear word or part of speech. In the connection table editing device to perform,
Information reading means for reading information stored in the connection table for morphological analysis ;
Based on the information read by the information reading means, the front word or part of speech is expanded to a row or column heading, and the rear word or part of speech is expanded to a column or row heading, and the intersection of each row and column. A table creation means for creating a table composed of rows and columns by arranging a connection rule relating to whether or not to connect a front word or part of speech to a back word or part of speech;
Display means for displaying a table of rows and columns the table producing means,
Input receiving means for receiving input from the user;
Based on information from the input receiving means, an editing operation means for editing the table created by the table creating means;
After editing the table by the editing calculation means, a duplication check means for judging whether or not a connection rule group possessed by a row or a column relating to the edited portion overlaps with a connection rule group possessed by another row or column;
When the duplication checking means determines that the connection rule group possessed by the row or column related to the edited portion overlaps with the connection rule group possessed by another row or column, one of the plurality of duplicate rows or columns or A merge means for adding other duplicate row or column headings to the column headings and deleting other row or column;
Format conversion means for converting a table composed of rows and columns into a format corresponding to the connection table for morphological analysis ;
Information writing means for writing the information after the format conversion to the connection table ;
Connection table editing apparatus, characterized in that it comprises a. When the table creation means creates a table based on the information from the information reading means, the words or parts of speech that partially match the character strings constituting the words or parts of speech used as the headings of the respective rows or columns are close to each other. The connection table editing device according to claim 1, wherein the connection table editing device is arranged so as to be arranged. The editing operation means is
A search for receiving search instruction information related to a word or part of speech related to the new addition input by the user when the word or part of speech is newly added, and searching the connection rule corresponding to the search instruction information from the table And
A copy unit that copies the connection rule instructed by the user among the connection rules searched by the search unit as a connection rule for words or parts of speech related to a new addition;
The connection table editing device according to claim 1, wherein the connection table editing device is provided. The editing operation unit, the connection according to any one of claims 1 to 3, characterized in that it has an editing protection unit for setting any part of the table consisting of rows and columns moiety and uneditable portions Edit Table editing device.

Images